view tests/test-fastannotate-revmap.py @ 46607:e9901d01d135

revlog: add a mechanism to verify expected file position before appending

If someone uses `hg debuglocks`, or some non-hg process writes to the .hg
directory without respecting the locks, or if the repo's on a networked
filesystem, it's possible for the revlog code to write out corrupted data.

The form of this corruption can vary depending on what data was written and how
that happened. We are in the "networked filesystem" case (though I've had users
also do this to themselves with the "`hg debuglocks`" scenario), and most often
see this with the changelog. What ends up happening is we produce two items
(let's call them rev1 and rev2) in the .i file that have the same linkrev,
baserev, and offset into the .d file, while the data in the .d file is appended
properly. rev2's compressed_size is accurate for rev2, but when we go to
decompress the data in the .d file, we use the offset that's recorded in the
index file, which is the same as rev1, and attempt to decompress
rev2.compressed_size bytes of rev1's data. This usually does not succeed. :)

When using inline data, this also fails, though I haven't investigated why too
closely. This shows up as a "patch decode" error. I believe what's happening
there is that we're basically ignoring the offset field, getting the data
properly, but since baserev != rev, it thinks this is a delta based on rev
(instead of a full text) and can't actually apply it as such.

For now, I'm going to make this an optional component and default it to entirely
off. I may increase the default severity of this in the future, once I've
enabled it for my users and we gain more experience with it. Luckily, most of my
users have a versioned filesystem and can roll back to before the corruption has
been written, it's just a hassle to do so and not everyone knows how (so it's a
support burden). Users on other filesystems will not have that luxury, and this
can cause them to have a corrupted repository that they are unlikely to know how
to resolve, and they'll see this as a data-loss event. Refusing to create the
corruption is a much better user experience.

This mechanism is not perfect. There may be false-negatives (racy writes that
are not detected). There should not be any false-positives (non-racy writes that
are detected as such). This is not a mechanism that makes putting a repo on a
networked filesystem "safe" or "supported", just *less* likely to cause
corruption.

Differential Revision: https://phab.mercurial-scm.org/D9952

from __future__ import absolute_import, print_function

import os
import tempfile

from mercurial import (
    pycompat,
    util,
)

from hgext.fastannotate import error, revmap

if pycompat.ispy3:
    xrange = range


def genhsh(i):
    return pycompat.bytechr(i) + b'\0' * 19


def gettemppath():
    fd, path = tempfile.mkstemp()
    os.close(fd)
    os.unlink(path)
    return path


def ensure(condition):
    if not condition:
        raise RuntimeError('Unexpected')


def testbasicreadwrite():
    path = gettemppath()

    rm = revmap.revmap(path)
    ensure(rm.maxrev == 0)
    for i in xrange(5):
        ensure(rm.rev2hsh(i) is None)
    ensure(rm.hsh2rev(b'\0' * 20) is None)

    paths = [
        b'',
        b'a',
        None,
        b'b',
        b'b',
        b'c',
        b'c',
        None,
        b'a',
        b'b',
        b'a',
        b'a',
    ]
    for i in xrange(1, 5):
        ensure(rm.append(genhsh(i), sidebranch=(i & 1), path=paths[i]) == i)

    ensure(rm.maxrev == 4)
    for i in xrange(1, 5):
        ensure(rm.hsh2rev(genhsh(i)) == i)
        ensure(rm.rev2hsh(i) == genhsh(i))

    # re-load and verify
    rm.flush()
    rm = revmap.revmap(path)
    ensure(rm.maxrev == 4)
    for i in xrange(1, 5):
        ensure(rm.hsh2rev(genhsh(i)) == i)
        ensure(rm.rev2hsh(i) == genhsh(i))
        ensure(bool(rm.rev2flag(i) & revmap.sidebranchflag) == bool(i & 1))

    # append without calling save() explicitly
    for i in xrange(5, 12):
        ensure(
            rm.append(genhsh(i), sidebranch=(i & 1), path=paths[i], flush=True)
            == i
        )

    # re-load and verify
    rm = revmap.revmap(path)
    ensure(rm.maxrev == 11)
    for i in xrange(1, 12):
        ensure(rm.hsh2rev(genhsh(i)) == i)
        ensure(rm.rev2hsh(i) == genhsh(i))
        ensure(rm.rev2path(i) == paths[i] or paths[i - 1])
        ensure(bool(rm.rev2flag(i) & revmap.sidebranchflag) == bool(i & 1))

    os.unlink(path)

    # missing keys
    ensure(rm.rev2hsh(12) is None)
    ensure(rm.rev2hsh(0) is None)
    ensure(rm.rev2hsh(-1) is None)
    ensure(rm.rev2flag(12) is None)
    ensure(rm.rev2path(12) is None)
    ensure(rm.hsh2rev(b'\1' * 20) is None)

    # illformed hash (not 20 bytes)
    try:
        rm.append(b'\0')
        ensure(False)
    except Exception:
        pass


def testcorruptformat():
    path = gettemppath()

    # incorrect header
    with open(path, 'wb') as f:
        f.write(b'NOT A VALID HEADER')
    try:
        revmap.revmap(path)
        ensure(False)
    except error.CorruptedFileError:
        pass

    # rewrite the file
    os.unlink(path)
    rm = revmap.revmap(path)
    rm.append(genhsh(0), flush=True)

    rm = revmap.revmap(path)
    ensure(rm.maxrev == 1)

    # corrupt the file by appending a byte
    size = os.stat(path).st_size
    with open(path, 'ab') as f:
        f.write(b'\xff')
    try:
        revmap.revmap(path)
        ensure(False)
    except error.CorruptedFileError:
        pass

    # corrupt the file by removing the last byte
    ensure(size > 0)
    with open(path, 'wb') as f:
        f.truncate(size - 1)
    try:
        revmap.revmap(path)
        ensure(False)
    except error.CorruptedFileError:
        pass

    os.unlink(path)


def testcopyfrom():
    path = gettemppath()
    rm = revmap.revmap(path)
    for i in xrange(1, 10):
        ensure(
            rm.append(genhsh(i), sidebranch=(i & 1), path=(b'%d' % (i // 3)))
            == i
        )
    rm.flush()

    # copy rm to rm2
    rm2 = revmap.revmap()
    rm2.copyfrom(rm)
    path2 = gettemppath()
    rm2.path = path2
    rm2.flush()

    # two files should be the same
    ensure(len({util.readfile(p) for p in [path, path2]}) == 1)

    os.unlink(path)
    os.unlink(path2)


class fakefctx(object):
    def __init__(self, node, path=None):
        self._node = node
        self._path = path

    def node(self):
        return self._node

    def path(self):
        return self._path


def testcontains():
    path = gettemppath()

    rm = revmap.revmap(path)
    for i in xrange(1, 5):
        ensure(rm.append(genhsh(i), sidebranch=(i & 1)) == i)

    for i in xrange(1, 5):
        ensure(((genhsh(i), None) in rm) == ((i & 1) == 0))
        ensure((fakefctx(genhsh(i)) in rm) == ((i & 1) == 0))
    for i in xrange(5, 10):
        ensure(fakefctx(genhsh(i)) not in rm)
        ensure((genhsh(i), None) not in rm)

    # "contains" checks paths
    rm = revmap.revmap()
    for i in xrange(1, 5):
        ensure(rm.append(genhsh(i), path=(b'%d' % (i // 2))) == i)
    for i in xrange(1, 5):
        ensure(fakefctx(genhsh(i), path=(b'%d' % (i // 2))) in rm)
        ensure(fakefctx(genhsh(i), path=b'a') not in rm)


def testlastnode():
    path = gettemppath()
    ensure(revmap.getlastnode(path) is None)
    rm = revmap.revmap(path)
    ensure(revmap.getlastnode(path) is None)
    for i in xrange(1, 10):
        hsh = genhsh(i)
        rm.append(hsh, path=(b'%d' % (i // 2)), flush=True)
        ensure(revmap.getlastnode(path) == hsh)
        rm2 = revmap.revmap(path)
        ensure(rm2.rev2hsh(rm2.maxrev) == hsh)


testbasicreadwrite()
testcorruptformat()
testcopyfrom()
testcontains()
testlastnode()